What is ZFS?
What Is ZFS?
ZFS is a filesystem and volume-management architecture designed to manage storage while emphasizing data integrity. Rather than treating the filesystem, logical storage and disk redundancy as completely separate layers, ZFS integrates these functions into a storage-pool architecture.
ZFS uses features such as end-to-end checksums, copy-on-write, storage pools, snapshots and redundant VDEV configurations to help detect corruption and protect stored data. Depending on the storage design, redundancy can be provided through mirrors or RAIDZ.
Why Is ZFS Different From a Traditional Filesystem?
Traditional storage architectures commonly separate physical disks, RAID, volume management and the filesystem into multiple layers. ZFS takes a more integrated approach.
Pooled Storage
Physical storage is organized into virtual devices that are used to build a ZFS storage pool.
Checksums
ZFS stores checksums for data and metadata so corruption can be detected when information is read.
Copy-on-Write
ZFS generally writes modified data to new locations rather than overwriting the existing blocks in place.
Integrated Redundancy
ZFS can use mirrors or RAIDZ configurations to provide redundancy across multiple storage devices.
Drives → VDEVs → Zpool → Datasets
One of the most important concepts in ZFS is understanding how physical storage devices become part of the overall storage pool. ZFS organizes storage differently from a conventional disk-and-partition model.
Drives
HDDs, SSDs or other supported storage devices provide the underlying physical capacity.
VDEVs
Drives are organized into virtual devices such as mirrors or RAIDZ groups.
Zpool
One or more VDEVs provide storage capacity to the ZFS storage pool.
Datasets
Filesystems and other logical storage objects can then be created from the pool's available capacity.
Zpools and VDEVs
What Is a Zpool?
A zpool is the storage pool from which ZFS allocates capacity. A pool can contain one or more VDEVs.
Learn About Zpools →What Is a VDEV?
A VDEV is a fundamental storage building block within a ZFS pool and can contain one or multiple drives.
Learn About VDEVs →What Is RAIDZ?
RAIDZ is a ZFS redundancy scheme that distributes data and parity across multiple storage devices. Different RAIDZ levels provide different amounts of parity and fault tolerance.
RAIDZ1
Uses single parity and can tolerate the loss of one device within the RAIDZ1 VDEV.
RAIDZ1 Guide →RAIDZ2
Uses double parity and can tolerate the loss of two devices within the RAIDZ2 VDEV.
RAIDZ2 Guide →RAIDZ3
Uses triple parity and can tolerate the loss of three devices within the RAIDZ3 VDEV.
RAIDZ3 Guide →ZFS Mirrors
ZFS does not require RAIDZ for redundant storage. A mirrored VDEV stores redundant copies of data across multiple devices. Mirrors can be attractive where performance, simpler redundancy groups or certain expansion strategies are priorities.
ZFS Is Designed to Detect Data Corruption
ZFS uses checksums to verify stored data and metadata. When redundant copies or parity are available, ZFS can use valid information from another location to repair certain detected errors rather than simply returning corrupted data.
Snapshots, Scrubs & Resilvering
ZFS Snapshots
Snapshots preserve a point-in-time view of a dataset without initially duplicating all of its stored data.
Snapshots Guide →ZFS Scrub
A scrub reads data in the pool and verifies it against ZFS checksums to identify integrity problems.
Scrub Guide →ZFS Resilvering
Resilvering reconstructs the required data on a device when restoring redundancy after certain drive events.
Resilvering Guide →Is ZFS the Same as RAID?
No. RAID generally describes techniques for combining multiple drives to provide performance, capacity and/or redundancy. ZFS is a broader storage architecture that includes filesystem functionality, storage pooling, checksumming, snapshots and integrated redundancy options such as mirrors and RAIDZ.
ZFS can therefore provide functionality traditionally associated with RAID, but ZFS and RAID should not be treated as interchangeable terms.
ZFS Is Not a Backup
ZFS redundancy and snapshots should not be treated as replacements for an independent backup. Mirrors and RAIDZ can protect against specified device failures, checksums can help detect corruption, and snapshots can preserve earlier filesystem states, but these protections do not cover every way data can be lost.
Important data should still be protected with an appropriate backup strategy that maintains independent copies according to the organization's recovery requirements.
Where Is ZFS Used?
ZFS FAQ
What does ZFS stand for?
ZFS originated as the Zettabyte File System, although the technology is now generally referred to simply as ZFS.
Is ZFS a filesystem or RAID?
ZFS is a broader filesystem and storage-management architecture that includes integrated redundancy options such as mirrors and RAIDZ.
What is a ZFS pool?
A ZFS storage pool, or zpool, combines the capacity provided by one or more VDEVs into a pool from which ZFS allocates storage.
What is RAIDZ?
RAIDZ is a ZFS parity-based redundancy scheme. RAIDZ1, RAIDZ2 and RAIDZ3 provide one, two and three parity levels respectively.
Can ZFS detect corrupted data?
ZFS uses checksums to detect integrity problems. When appropriate redundancy is available, it can repair certain detected errors from valid redundant data.
Does ZFS require RAIDZ?
No. ZFS pools can use several VDEV layouts, including mirrored configurations. The appropriate design depends on capacity, performance and redundancy requirements.
Are ZFS snapshots backups?
A snapshot by itself should not be considered an independent backup because it normally remains dependent on the underlying storage system.
What happens when a ZFS drive fails?
The result depends on the VDEV configuration and remaining redundancy. In a redundant configuration, a failed device can generally be replaced and the required data resilvered onto the replacement.